Sentiment Analysis Dashboard

This project provides a sentiment analysis dashboard built using Streamlit and RoBERTa, designed to analyze Reddit posts from selected subreddits. It visualizes sentiment trends, distributions, and key insights from posts.

Features

• Fetch Reddit Data: Retrieves posts from specified subreddits using the Reddit API.
• Sentiment Analysis: Leverages the siebert/sentiment-roberta-large-english model to classify post titles into positive, negative, or neutral sentiments.
• Interactive Dashboard:
  • Word cloud visualization for post titles.
  • Sentiment label distribution chart.
  • Rolling sentiment trends with hourly aggregation.
  • Insights into the most positive and negative posts.
• Customizable Filters:
  • Select subreddits.
  • Adjust the rolling average window.
  • Specify the number of posts to fetch per subreddit.

Requirements

• Python 3.8+
• Required libraries:
  • praw
  • transformers
  • pandas
  • matplotlib
  • seaborn
  • wordcloud
  • plotly
  • streamlit

Setup

1. Clone the Repository:

   git clone https://github.com/your_username/sentiment_analysis.git
   cd sentiment_analysis

2. Install Dependencies: It is recommended to use a virtual environment:

   python -m venv sentiment_env
   source sentiment_env/bin/activate  # On Windows: sentiment_env\Scripts\activate
   pip install -r requirements.txt

3. Set Up Reddit API Credentials:

   • Create an app in your Reddit account: Reddit App Settings
   • Update the client_id, client_secret, and user_agent in the script.

4. Run the Dashboard:

   streamlit run sentiment_dashboard.py

Usage

• Select Subreddits: Choose one or more subreddits to analyze.
• Adjust Parameters: Modify the rolling average window and the number of posts to fetch.
• Visualize Results:
  • View sentiment trends and insights.
  • Explore raw data with direct links to Reddit posts.

File Structure

• sentiment_dashboard.py: Main script for the Streamlit dashboard.
• requirements.txt: List of dependencies.
• Other Python scripts and data files (if applicable).

Visualizations

Word Cloud

• A visualization of the most frequently used words in post titles.

Sentiment Label Distribution

• Bar chart showing the count of positive, negative, and neutral posts.

Rolling Sentiment Trend

• Line chart displaying sentiment trends aggregated hourly.

Known Issues

• Some posts may be excluded if the title exceeds the RoBERTa model's token limit.
• Errors can occur if Reddit API credentials are invalid.

Contributions

Contributions are welcome! If you'd like to improve this project, feel free to fork the repository, make changes, and submit a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for more details.

---

Contact

If you have any questions or feedback, please reach out through GitHub or email.